import os
import openai
import streamlit as st
from dotenv import load_dotenv
import boto3
import base64
import datetime
import json

load_dotenv()
OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")
client = openai.OpenAI(api_key=OPENAI_API_KEY)

dynamodb = boto3.resource(
    'dynamodb',
    aws_access_key_id=os.getenv('AWS_ACCESS_KEY_ID'), 
    aws_secret_access_key=os.getenv('AWS_SECRET_ACCESS_KEY'),
    region_name='ap-northeast-2'  # 원하는 리전으로 변경
)
table = dynamodb.Table('ChatbotConversations')

#original_allowed_usernames = os.getenv('ALLOWED_NAMES')
#allowed_usernames = json.loads(original_allowed_usernames)

original_allowed_usernames = os.getenv('ALLOWED_NAMES')
original_allowed_usernames = original_allowed_usernames.strip()

# JSON 파싱을 수행합니다.
allowed_usernames = json.loads(original_allowed_usernames)

def is_valid_username(username):
    return username in allowed_usernames

st.title(':male-teacher: AI ROBLOX Tutor - Feedback')
st.header(':two: Tutor for Feedack: Get feedbacks and suggestions on your own Roblox game codes!')

user_id = st.text_input(label='Assigned ID.')


if is_valid_username(user_id):
    st.write(f'ID: :violet[{user_id}]')

    purpose = st.text_input(
        label='Write the purpose of the code that you want to get feedback on.',
        placeholder='purpose of your code'
    )

    st.write(f'Purpose: :violet[{purpose}]')

    eval = st.text_input(
        label='Write the code that you want to get feedback on.',
        placeholder='copy and paste syntax-based code script'
    )

    st.write(f'Codes: :violet[{eval}]')

    query2 = "purpose: " + purpose + ". texted codes: " + eval

    query_2 = 'AT_TUTOR_2: ' + query2

    # 이미지 입력
    image_uploaded2 = st.file_uploader("OPTIONAL: Upload an image of the code that you want to get feedback on", type=["png", "jpg", "jpeg"])

    def evaluator_persona_query(eval, purpose, image_data=None):
        import openai
        openai.api_key = OPENAI_API_KEY

        VISION_PROMPT_MESSAGES = [
            {
                "role": "system",
                "content": "Your role: You are an AI tutor providing corrective and explanatory feedback on students' Roblox game scripts. Your goal is to help learners understand computational thinking concepts through meaningful, targeted feedback. Context: - The learner is coding in Lua within the Roblox game development environment. - The learner is submitting a code snippet along with a brief description of its intended function. - The tutor provides structured feedback based on **computational thinking (CT) concepts** such as: - **Algorithm & Procedures:** Does the code follow structured logic? - **Automation & Efficiency:** Are there redundant lines or unnecessary loops? - **Debugging & Error Handling:** Does the code have syntax/logical errors? - **Modularity & Reusability:** Are functions used effectively? User Input: - A code snippet written in Lua for a game feature. - A description of what the learner **intends** the code to do. Expected AI Response: 1. **Confirm understanding** by summarizing what the code is supposed to do. 2. **Identify and correct errors** (if any) with explanations: - Highlight incorrect syntax or logic errors. - Suggest alternative, optimized code structures. 3. **Provide explanations** for computational thinking concepts that apply to this code. 4. **Suggest improvements** based on best practices in coding and game design. 5. **Encourage reflection** by asking the learner to predict how changes will affect gameplay." 
                ##CSTA & ISTE (2011). Computational Thinking in K-12 education – teacher resources, 2nd edition, CSTA & ISTE
            },
            {"role": "user", "content": "Read and find the meaning of the code in the image and text and evaluate the code. purpose of the code: "+ purpose + "code to be analysed: " + eval},
        ]

        if image_data is not None:
            encoded_image = base64.b64encode(image_data).decode("utf-8")
            VISION_PROMPT_MESSAGES.append({"role": "user", "content": encoded_image})

        params = {
            "model": "gpt-4o-mini",
            "messages": VISION_PROMPT_MESSAGES,
            "max_tokens": 3000, # 1024,
        }

        trimmed_answer = ""  # trimmed_answer 초기화
        try:
            full_answer = openai.chat.completions.create(**params)
            trimmed_answer = full_answer.choices[0].message.content
        except Exception as e:
            print(f"Error: {e}")
            trimmed_answer = f"Error: {e}"  # 오류 메시지를 trimmed_answer에 할당

        return trimmed_answer

    # 버튼 클릭
    button2 = st.button(':sparkles: suggestions :sparkles:')

    def save_message(user_id, message, timestamp=None):
        if timestamp is None:
            timestamp = datetime.datetime.now()

        # datetime 객체를 DynamoDB 지원 타입으로 변환
        if isinstance(timestamp, datetime.datetime):
            timestamp = int(timestamp.timestamp())  # Unix 타임스탬프(초 단위 정수)로 변환

        table.put_item(
            Item={
                'UserID': user_id,
                'Timestamp': timestamp,
                'Message': message
            }
        )


    if button2:
        eval = eval
        save_message(user_id, query_2) 

        if image_uploaded2 is not None:
            image_data = image_uploaded2.read()
            answer = evaluator_persona_query(eval, purpose, image_data)
        else:
            answer = evaluator_persona_query(eval, purpose)
        
        answer_2 = 'AT_TUTOR_2: ' + answer
        save_message(user_id, answer_2)
        st.write(f'{answer}')

else:
    st.warning("Invalid username. Please enter a valid username.")